AU2021107643A4 - Method, system, device, computer device and storage medium for elevator fault prediction - Google Patents

Abstract

The present application relates to a method, a system, a device, a computer device, and a storage medium for elevator fault prediction. The method includes: acquiring historical fault data and historical state data corresponding to the historical fault data of an elevator as sample data, the historical fault data including fault data of multiple fault types; inputting the sample data into a fault prediction model to obtain a fault correlation table of sample feature parameters of the sample data and individual fault types; sending the fault correlation table to an edge server. The edge server is configured to obtain state data and fault data of the elevator, determine current sample feature parameters according to the state data and the fault data, and determine a probability of occurrence of any fault type according to the current sample feature parameters and the fault correlation table. Adopting this method can reduce the delay and has the timeliness, while ensuring the improvement of the computing ability of the elevator side. acquiring historical fault data and historical state data corresponding to the historical fault data of an elevator as sample data, the historical fault data S302 including fault data of multiple fault types inputting the sample data into a fault prediction model to obtain a fault correlation table of sample feature parameters of the sample data and individual fault types, the fault correlation table recording a correlation S304 weight between individual sample feature parameters and a fault type when the fault type occurs sending the fault correlation table to an edge server, the edge server being configured to obtain state data and fault data of the elevator, determine current sample feature parameters according to the state data and the fault S306 data, and determine a probability of occurrence of any fault type according to the current sample feature parameters and the fault correlation table FIG. 3

Description

METHOD, SYSTEM, DEVICE, COMPUTER DEVICE AND STORAGE MEDIUM FOR ELEVATOR FAULT PREDICTION TECHNICAL FIELD

[0001] The present application relates to a field of elevator fault handling, in particular to a method, an elevator fault prediction system, a device, a computer device, and a storage medium for elevator fault prediction.

BACKGROUND

[0002] In recent years, with the improvement of people's living standards, elevators have gradually become essential equipment in residential buildings, office buildings, shopping malls and other buildings. The safety of the elevators directly affects the lives of people who take the elevators. Therefore, finding possible elevator faults in time and eliminating them in advance are the focus of elevator research. There are mainly two conventional elevator fault diagnosis methods. As shown in FIG. 1, an elevator fault diagnosis method uses a main control board to determine whether the fault occur according to the real-time data during an operation of the elevator, and another elevator fault diagnosis method uses a main control board to transmit the operating data to a central server, and then uses a specific process and a threshold of the central server to determine whether the elevator is faulty.

[0003] However, although in the first diagnosis method, the fault can be diagnosed via the main control board without relying on the central server, due to the limitation of the storage capacity and computing ability of the edge device, the main control board cannot cache a large amount of historical data, nor can perform very complex logic analysis, which lead to the lack of repeated learning and recurrence of large amounts of data, and therefore causes the main control board to only run relatively simple judgment logic. In order to implement complex calculation and analysis, in the second diagnosis method, data is transmitted to the server, and a variety of logical judgments can be realized through offline statistics and analysis, but this method needs to rely on the central server and has low timeliness.

[0004] Therefore, the conventional fault diagnosis method has a problem of incompatibility between the timeliness of diagnosis and the ability of logical judgment.

SUMMARY

[0005] Accordingly, with respect to the problems of incompatibility between the timeliness of the diagnosis and the ability of the logical judgment in the above fault diagnosis methods, it is necessary to provide a method, a system, and a device, a computer device, and a storage medium for elevator fault prediction.

[0006] A method for elevator fault prediction includes:

[0007] acquiring historical fault data and historical state data corresponding to the historical fault data of an elevator as sample data, the historical fault data including fault data of multiple fault types;

[0008] inputting the sample data into a fault prediction model to obtain a fault correlation table of sample feature parameters of the sample data and individual fault types, the fault correlation table recording a correlation weight between individual sample feature parameters and a fault type when the fault type occurs;

[0009] sending the fault correlation table to an edge server, wherein the edge server is configured to obtain state data and fault data of the elevator, determine current sample feature parameters according to the state data and the fault data, and determine a probability of occurrence of any fault type according to the current sample feature parameters and the fault correlation table.

[0010] In one of the embodiments, before inputting the sample data into the fault prediction model, the method further includes:

[0011] determining a target fault and multiple target fault types of the target fault, and filtering out target historical fault data and target historical state data corresponding to the target fault and the target fault types from the historical fault data and the historical state data, as the sample data.

[0012] The inputting the sample data into the fault prediction model to obtain the fault correlation table of the sample feature parameters of the sample data and the individual fault types includes:

[0013] using fault codes of the individual target fault types as target fault codes, and dividing the sample data into a positive sample whose fault code is a sample of the target fault code and a negative sample whose fault code is not a sample of the target fault code;

[0014] determining multiple sample features that characterize the sample data, obtaining sample feature parameters of individual sample features through statistical processing of the positive sample and the negative sample, and performing correlation analysis on the sample feature parameters and the target fault codes, to obtain the fault correlation table.

[0015] In one of the embodiments, the determining the target fault and multiple target fault types of the target fault includes:

[0016] acquiring the number of individual fault codes in the historical fault data; and

[0017] filtering out a fault code with the number of occurrences exceeding a set number threshold from the individual fault codes, using a fault type corresponding to the fault code as the target fault type.

[0018] In one of the embodiments, the sample feature parameters include fault parameters. The fault parameters include the fault codes and the number of fault codes.

[0019] The performing correlation analysis on the sample feature parameters and the target fault codes to obtain the fault correlation table includes:

[0020] dividing the number of the fault codes into at least two categories;

[0021] according to the number of the individual fault codes and the divided categories, performing correlation analysis on the number of the individual fault codes and the target fault codes by a chi-square test method, to obtain a correlation weight between the target fault code and the number of the individual fault codes before the occurrence of a fault type corresponding to the target fault code.

[0022] In one of the embodiments, the sample feature parameters further include state parameters. The state parameters include instantaneous indexes and index values of the instantaneous indexes.

[0023] The performing correlation analysis on the sample feature parameters and the target fault codes to obtain the fault correlation table further includes:

[0024] determining multiple instantaneous indexes, performing statistical processing on the historical state data according to a preset time window length, to obtain index values of individual instantaneous indexes;

[0025] discretizing the index values of the individual instantaneous indexes into at least two categories, and according to the discretized categories and the index values of the individual instantaneous indexes, performing correlation analysis on the index values of the individual instantaneous indexes and the target fault codes by a chi-square test method, to obtain a correlation weight between the target fault code and the index values of the individual instantaneous indexes before the occurrence of a fault type corresponding to the target fault code.

[0026] A system for elevator fault prediction includes an edge server and a central server that communicate through a network.

[0027] The central server is configured to acquire historical fault data and historical state data corresponding to the historical fault data of an elevator as sample data, the historical fault data including fault data of multiple fault types; to input the sample data into a fault prediction model to obtain a fault correlation table of sample feature parameters of the sample data and individual fault types, the fault correlation table recording a correlation weight between individual sample feature parameters and a fault type when the fault type occurs; to sending the fault correlation table to the edge server.

[0028] The edge server is configured to acquire state data and fault data of the elevator, determine current sample feature parameters according to the state data and the fault data, and determine a probability of occurrence of any fault type according to the current sample feature parameters and the fault correlation table.

[0029] In one of the embodiments, the edge server is further configured to acquire a life index of an elevator operation and an index value of the life index, and determine a probability of occurrence of any fault type according to the index value of the life index, the current sample feature parameters and the fault correlation table.

[0030] An elevator fault prediction device includes:

[0031] a data acquisition module configured to acquire historical fault data and historical state data corresponding to the historical fault data of an elevator as sample data, the historical fault data including fault data of multiple fault types;

[0032] a correlation table generation module configured to input the sample data into a fault prediction model to obtain a fault correlation table of sample feature parameters of the sample data and individual fault types, the fault correlation table recording a correlation weight between the individual sample feature parameters and a fault type when the fault type occurs; and

[0033] a correlation table sending module configured to send the fault correlation table to an edge server. The edge server is configured to acquire state data and fault data of the elevator, determine current sample feature parameters according to the state data and the fault data, and determine a probability of occurrence of any fault type according to the current sample feature parameters and the fault correlation table.

[0034] A computer device includes a memory storing a computer program and a processor. When executing the computer program, the processor implements steps of:

[0035] acquiring historical fault data and historical state data corresponding to the historical fault data of an elevator as sample data, the historical fault data including fault data of multiple fault types;

[0036] inputting the sample data into a fault prediction model to obtain a fault correlation table of sample feature parameters of the sample data and individual fault types, the fault correlation table recording a correlation weight between individual sample feature parameters and a fault type when the fault type occurs; and

[0037] sending the fault correlation table to an edge server, wherein the edge server is configured to obtain state data and fault data of the elevator, determine current sample feature parameters according to the state data and the fault data, and determine a probability of occurrence of any fault type according to the current sample feature parameters and the fault correlation table.

[0038] A computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the following steps are implemented:

[0039] acquiring historical fault data and historical state data corresponding to the historical fault data of an elevator as sample data, the historical fault data including fault data of multiple fault types;

[0040] inputting the sample data into a fault prediction model to obtain a fault correlation table of sample feature parameters of the sample data and individual fault types, the fault correlation table recording a correlation weight between individual sample feature parameters and a fault type when the fault type occurs; and

[0041] sending the fault correlation table to an edge server, wherein the edge server is configured to obtain state data and fault data of the elevator, determine current sample feature parameters according to the state data and the fault data, and determine a probability of occurrence of any fault type according to the current sample feature parameters and the fault correlation table.

[0042] In the above method, system, device, computer device and storage medium for elevator fault prediction, according to the historical fault data and corresponding historical state data of the elevator, the central server performs the model training in the fault prediction model, to obtain the fault correlation table of the sample feature parameters of the sample data and the individual fault types. The fault correlation table records the correlation weight between the individual sample feature parameters and the fault types when the individual fault types occur. The central server sends the fault correlation table to the edge server, so that the edge server obtains the fault data and the state data of the elevator, determine the current sample feature parameters according to the fault data and state data, and then can directly determine the probability of occurrence of any fault type according to the sample feature parameters and the fault correlation table, thereby realizing the prediction of elevator fault. As such, the edge server can realize the storage and logical calculation of the operating data of the elevator, perform fault prediction, realize complex on-site calculations and on-site storage of large amounts of data, without being restricted by the performance of the main control board of the elevator, and without returning the acquired data to the central server. As such, the storage and calculation can be realized at the edge node close to the elevator. While ensuring the improvement of the computing ability of the elevator side, the delay is reduced, and the timeliness is high. In order to distribute and store the data, and diverge the operation of the logical operations, the central server only needs to perform the model training, which greatly reduces the pressure on the central server. In addition, the large-scale edge-side storage and calculation can be achieved without purchasing a large number of central servers. There is no need to equip the elevator with large-capacity storage units and processing chips, the programs are directly deployed on the edge cloud provided by operators, capacity and performance can be improved, and program improvements and upgrades are also easier.

BRIEF DESCRIPTION OF THE DRAWINGS

[0043] FIG. 1 is a scene view of a conventional elevator fault diagnosis method according to an embodiment.

[0044] FIG. 2 is an application scene view of a method for elevator fault predication according to an embodiment.

[0045] FIG. 3 is a schematic flowchart of a method for elevator fault predication according to an embodiment.

[0046] FIG. 4 is a schematic flowchart of steps of generating a fault correlation table according to an embodiment.

[0047] FIG. 5 is a flowchart of a sample feature parameter correlation analysis according to an embodiment.

[0048] FIG. 6 is a schematic view of a generated form of sample features according to an embodiment.

[0049] FIG. 7 is a schematic view of a system for elevator fault prediction according to an embodiment.

[0050] FIG. 8 is a block diagram of a device for elevator fault prediction according to an embodiment.

[0051] FIG. 9 is an inner structural diagram of a computer device according to an embodiment.

DETAILED DESCRIPTION OF THE EMBODIMENTS

[0052] In order to make the purpose, solutions, and advantages of the present application clearer, the present application will be further illustrated in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application, and are not used to limit the present application.

[0053] A method for elevator fault prediction according to the present application can be applied to an application environment as shown in FIG. 2. By adopting 5G edge cloud computing technology, an edge server 202 is deployed in a machine room adjacent to an elevator side. In a central server 204, uses historical fault data and corresponding historical state data of an elevator are used as sample data, the sample data is input into a fault prediction model for model training to obtain a fault correlation table of sample feature parameters of the sample data and individual fault types. In the fault correlation table, correlation weights between individual sample feature parameters and a fault type are recorded when the fault type occurs. The central server 204 sends the fault correlation table to the edge server 202, so that the edge server 202 obtains the fault data and state data of the elevator, determines current sample feature parameters according to the fault data and state data, and then can directly determine the probability of occurrence of any fault type according to the sample feature parameters and the fault correlation table, so as to realize the prediction of the elevator faults. As such, the storage and logical operation of operating data of the elevator can be realized in the edge server 202 to realize the fault diagnosis without returning the acquired data to the central server 204. Through lowering down the operation logic to an edge site or a place close to the edge, communication input and output (10) are reduced, program upgrades are simplified, and there is no need to face a huge number of devices, but only edge nodes of the cloud, thereby increasing computing ability and reducing latency. The edge server 202 and the central server 204 communicate through a network. The edge server 202 and the central server 204 may be implemented by independent servers or a server cluster composed of multiple servers.

[0054] In an embodiment, as shown in FIG. 3, a method for elevator fault prediction is provided. Taking the method applied to the central server 204 in FIG. 2 as an example for illustration, the method includes steps of S302 to S306 as follows.

[0055] At Step S302, historical fault data and historical state data corresponding to the historical fault data of the elevator are acquired as sample data. The historical fault data includes fault data of multiple fault types.

[0056] The historical fault data indicates the relevant data for each fault of the elevator. For example, the historical fault data may include the elevator model, the time of occurrence of the fault, the fault type/fault description information, the fault code, the fault level, the risk level, etc. As shown in Table 1 below, it is an example of the historical fault data. Table 1 Elevator Risk level Fault type/Fault description Fault code model Fault level

Lowerefficiency Locking fault during opening CA E or closing the doors has d7 (certainly occur) ocuednhps occurred in the past

CA A2 Emergency stop Momentary interruption fault 30 (certainly occur) of hall door lock circuit

CA C1 Trapping people Blocked fault during opening 82 (top risk) and closing the doors continuous action fault of the CALowerefficiency ORS during closing the main d8 (certainly occur) door

CA C2 Trapping people fault of the inverter door bO (top risk) machine

[0057] The historical state data indicates the operating state data before the elevator failure. For example, the historical state data can be the time from issuing the elevator door opening command to turning off the car door switch, the time from issuing the door opening command to turning off the door lock, the average operating current during opening the doors, and the average operating current during closing the doors, door opening time, etc.

[0058] In practical applications, correlation analysis can be performed based on elevator models. That is, for each elevator model, corresponding historical fault data and historical state data are obtained, and a corresponding fault correlation table is generated for each elevator model. Since the correlation analysis is based on the elevator model, a sample data threshold can also be set. The sample data threshold can be 500. When all the sample data of a certain elevator model exceeds the sample data threshold, the correlation result for the elevator model between fault codes of individual fault types and other faults or state data are calculated. If it is lower than the sample data threshold, the number of samples is too small, the result of the correlation analysis is greatly affected by noise, and the accuracy of the result of the analysis will also be affected. Therefore, the sample data of this elevator model may not be processed.

[0059] Further, in an embodiment, after obtaining the historical fault data and the historical state data corresponding to the historical fault data of the elevator, the method further includes: determining a target fault and multiple target fault types of the target fault, and filtering out the target historical fault data and the target historical state data corresponding to the target fault and the target fault types from the historical fault data and the historical state data as sample data.

[0060] The target fault indicates a fault that needs to be predicted, and the target fault in the present application may include elevator stop faults and door machine faults.

[0061] Specifically, after the target fault is selected, multiple fault types corresponding to the target fault are determined, and each fault type has a corresponding fault code. Therefore, the historical fault data corresponding to the fault code of the fault type of the target fault can be filtered out from the historical fault data according to the fault code as the target fault data, and then the target state data corresponding to the target fault data is filtered out from the historical state data. The target fault data and the target state data are used as the sample data.

[0062] In practical applications, the door machine fault can be determined by matching the fault code. For the elevator stop fault, the elevator stop records can be used to find out the historical fault data with the fault level of category A/B/C within the set time (such as 5 minutes) before and after each elevator stop fault, and select the fault record with OrderNO as the target fault of the elevator stop fault prediction.

[0063] Further, after the sample data is obtained, it is further necessary to filter the sample data to filter out the sample data that meets the preset conditions. More specifically, the data in the sample data whose state parameters do not meet the parameter threshold requirements can be filtered, for example, the data with a speed greater than 2000 or a floor code value greater than 200 can be filtered out; and/or, the data whose fault time is less than the time threshold in the sample data can be filtered out, for example, a determination of whether the fault is resolved within 5 minutes is performed, and if so, the fault is filtered out; and/or, the fault data that has been in the fault state in the sample data is filtered, for example, a determination of whether there are other fault data whose IsMaintained field value is 1 within half an hour before the occurrence of the fault is performed. If there is, the fault is filtered; and/or, the sample data for which the difference between the set parameters before and after the preset time period does not meet the difference threshold is filtered, for example, by comparing the state data in several data packages before and after the day, the data satisfying any one of the following conditions are filtered out: the difference between RunTimes is greater than the first set times (such as 3000 times), the difference between RunTotalTime is greater than the set time (such as 24x3600 seconds), and the difference between DoorTimes is greater than the second set times (such as 5000 times).

[0064] Through the above filtering processing, the data that is not the real fault is filtered out, and the obtained filtered data is the effective sample data, so that when the filtered sample data is sequentially input into the fault prediction model for correlation analysis, the accuracy of the analysis result can be improved.

[0065] At Step S304, the sample data is input into the fault prediction model to obtain a fault correlation table of the sample feature parameters of the sample data and individual fault types. The fault correlation table records the correlation weight between the individual sample feature parameters and a fault type when the fault type occurs.

[0066] In a specific implementation, after determining the target fault to be predicted and multiple target fault types of the target fault, the fault codes of the individual target fault types of the target fault to be predicted can be used as the target fault codes in turn, and the sample data can be divided into that a positive sample whose fault code are the target fault code and a negative sample whose fault code is not the target fault code. Multiple sample features used to characterize the sample data are determined. The feature parameters of the individual sample features are obtained through statistical processing of the positive and negative samples. The correlation analysis is performed on the sample feature parameters and the target fault code, to obtain the correlation weight between the sample feature parameters and a fault type when the fault type occurs, which is recorded as the fault correlation table.

[0067] After determining a target fault code, all sample data can be divided into the following two types: 1), the fault code is the target fault code; 2), the fault code is not the target fault code. Therefore, the distribution state of the obtained target fault codes is a two-category distribution. In order to characterize the correlation between the distribution of the target fault codes and other sample feature parameters, an appropriate reference value needs to be selected. Therefore, a chi-square test method can be used, in which the p-value, degree of freedom and test value of the chi-square distribution are used to comprehensively determine the correlation between the individual sample feature parameters and the target fault codes.

[0068] At Step S306, the fault correlation table is sent to the edge server. The edge server is used to obtain the state data and fault data of the elevator, determine the current sample feature parameters according to the state data and the fault data, and determine the probability of occurrence of any fault type according to the current sample feature parameters and the fault correlation table.

[0069] In a specific implementation, the machine learning is performed in the central server through the fault prediction model. After the fault correlation table of the sample feature parameters and the individual fault types is obtained, the fault correlation table can be sent to the edge server. As such, the edge server can obtain the state data and the fault data of the elevator in real time, and then determine the current sample feature parameters according to the state data and the fault data, and determine the probability of occurrences of the individual each fault types at the current sample feature parameters according to the correlation weight between the individual sample feature parameters when the individual fault type occurs.

[0070] In the above method for elevator fault prediction, in the central server, based on the historical fault data and the corresponding historical state data of the elevator, the model training in the fault prediction model is performed to obtain the fault correlation table of the sample feature parameters of the sample data and the individual fault types. The fault correlation table records the correlation weight between the individual sample feature parameters and the fault types when the individual fault type occurs. The central server sends the fault correlation table to the edge server, so that the edge server can obtain the fault data and state data of the elevator, and determine the current sample feature parameters according to the fault data and the state data, and then can directly determine the probability of occurrence of any fault type according to the sample feature parameters and the fault correlation table, thereby realizing the prediction of the elevator fault. As such, the edge server can realize the storage and logical calculation of the operating data of the elevator, perform fault prediction, realize complex on-site calculations and on-site storage of large amounts of data, without being restricted by the performance of the main control board of the elevator, and without returning the obtained data to the central server. As such, the storage and calculation can be realized at the edge node close to the elevator. While ensuring the improvement of the computing ability of the elevator side, the delay is reduced, and the timeliness is high. The data is distributed and stored, and the logical operations are run divergently, so that the central server only needs to perform the model training, which greatly reduces the pressure on the central server. In addition, the large-scale edge-side storage and calculation can be achieved without purchasing a large number of central servers. There is no need to equip the elevator with large-capacity storage units and processing chips, programs are directly deployed on the edge cloud provided by operators, the capacity and performance can be improved, and it is also easier to improve and upgrade the programs.

[0071] In an embodiment, as shown in FIG. 4, the above step S304 specifically includes the following steps of S402 to S404.

[0072] At step S402, the fault codes of the individual target fault type are used as the target fault codes respectively, and the sample data is divided into the positive sample and the negative sample. The positive sample indicates that the fault code is a sample of the target fault code, and the negative sample indicates that the fault code is not a sample of the target fault code.

[0073] At Step S404, multiple sample features that characterize the sample data are determined, the sample feature parameters of the individual sample features are obtained through statistical processing of the positive sample and the negative sample, and the correlation analysis is performed on the sample feature parameters and the target fault codes, to obtain the fault correlation table.

[0074] The sample feature parameters include the fault parameters and the state parameters. The fault parameters include the fault codes and the number of fault codes.

[0075] In a specific implementation, the correlation analysis of the sample data includes a fault part correlation analysis and a state part correlation analysis. The fault part correlation analysis refers to, when a certain fault X occurs, analyzing whether other faults such as Y that has occurred within a certain period of time before the occurrence of the certain fault X occur or not, how many times they have occurred, and the correlation thereof with the fault X. The state part correlation analysis refers to, when a certain fault X occurs, analyzing the statistical results of various states in a certain period of time before the occurrence of the certain fault X, such as the mean/variance etc. of the elevator door opening and closing time within 0 minutes to 5 minutes before the occurrence of the fault X, the correlation thereof with the fault X. Therefore, firstly, based on the elevator model, the fault codes of the target fault types at this elevator model can be used as the target fault codes in sequence, recorded in the combination form of "elevator model a-fault code X", and the "positive samples" and "negative samples" corresponding to the combination can be generated. As such, the positive sample indicates a sample for which the elevator model is equal to a, and the fault code is the target fault code X. The negative sample indicates a sample for which the elevator model is equal to a, and the fault code is not equal to the target fault code X. After that, the sample feature parameters of the individual sample features are processed statistically, and the correlation analysis is performed on the sample feature parameters and the target fault codes by the chi-square test method, to obtain the fault correlation table.

[0076] For example, taking the correlation analysis on the number of faults and the target fault code as an example, as shown in Table 2 below, in the example, the chi-square test is used to determine the correlation between the occurrence of the fault Y and the occurrence of the target fault X. The data in the first column in the table indicates three grades according to the number of target faults f2. The data in the second column in the table indicates, within 30 minutes before the occurrence of the target faults f2, the number of occurrences of the three cases where the fault 30 occurs 0 times, 1 time, and more than 1 time respectively, which is obtained through statistical processing of the positive sample. The data in the third column indicates, within 30 minutes before the occurrence of the non-target fault, the number of occurrences of the three cases where the fault 30 occurs 0 times, 1 time, and more than1 time respectively, which is obtained through statistical processing of the negative sample. The fifth row and the sixth column indicate the total number of statistics in the corresponding column or row. Table 2 the number of the non-target faults 30 within 30 target fault (f2) total number fault minutes before fault

0 33423 768112 801535

1 21 710 731

>1 21 862 883

total number 33465 769684 803149

[0077] In this table, a combined statistics are made according to whether the fault is the target fault (f2), and the number of occurrences of the faults with a fault code "30" within minutes before the fault. The number of occurrences of the fault code "30" is divided into 3 grades, that is, divided into 0 times, 1 time and more than or equal to 1 time. Firstly, an assumption to be tested of "the number of occurrences of the fault 30 within the 30 minutes before the fault is not related to whether the fault is the target fault f2" is made. Based on this assumption, the target faults and the non-target faults should be statistically processed independently, that is, the ratio of target faults/non-target faults should be basically the same in the three grades of 0, 1, and >1, which are equal to the ratio of 33465/768964z0.0435. Next, the degree of inconsistency between the actual statistical processing results and the assumptions is calculated.

[0078] For the grid of "0-target fault", according to the assumption of independent statistical processing, its value should be 801535*(33465/803149)=33397. The deviation of this value from the actual data is calculated as: (33423-33397)2/33397=0.02. The same method is used to calculate the deviation for the "0-non-fault target" to be 0.0008, the deviation for "1-target fault" to be 3.857, the deviation for "1-non-target fault" to be 0.1155, the deviation for ">1-target fault" to be 6.919, and the deviation for "1-non-target fault" to be 0.302, and thus the total deviation is calculated to be 10.158. Since the number of the faults 30 within the 30 minutes before the fault is divided into 3 categories, and the target faults/non-target faults are divided into two categories, the degree of freedom is (3-1)*(2-1)=2. The chi-square distribution when the degree of freedom is 2 is looked up, to obtain that the p value corresponding to 10.158 is 0.006. This means that when the assumption of independent statistical processing is established, the probability of the occurrence of the statistical result is only 0.006, and thus the assumption can be rejected, and then the distributions of the faults 30 and f2 can be determined to be correlated, and the correlation weight between the faults 30 and the target fault f2 can be calculated according to the statistical results in Table 2. Similarly, the correlation weight between the state parameters and the target faults can be obtained through the statistical result of the state parameters before the occurrences of the target faults. By analogy, the fault correlation table of the individual sample feature parameters and the individual target faults can be obtained.

[0079] The fault correlation table can be expressed as the "elevator model-fault code-sample feature parameter correlation table", which records the name of and correlation weight between the individual sample feature parameters under a condition of each combination of "elevator model-fault code" and that various sample feature parameters are sorted from high to low in terms of the correlation. Further, the fault correlation table can be periodically updated regularly to adapt to changes in the state of the elevator.

[0080] Refer to FIG. 5, which is a flowchart of sample feature parameter correlation analysis. In the historical data analysis stage, that is, in the central server, the historical fault data and the historical state data are obtained as historical samples (i.e., the sample data), the sample data is divided into the positive sample and the negative sample for correlation analysis to obtain a fault code-elevator model-parameter (i.e., the sample feature parameter) correlation table, which records the name of the individual sample feature parameters and the correlation weight with the target faults. In the real-time data query stage, that is, in the edge server, when predicting whether a certain fault code is likely to occur, the predicted time, elevator model, and fault code can be input. The state data and fault data of the elevator within a period of time before the prediction time are obtained from the sensors, counters or the like mounted on the elevator. From the state data and the fault data, the statistical processing is performed to obtain the current sample feature parameters. The probability of occurrence of the predicted fault code is determined, according to the current sample feature parameters and the fault code-elevator model-parameter (i.e., the sample feature parameter) correlation table.

[0081] In this embodiment, the sample data is divided into the positive sample and the negative sample by sequentially using the individual fault codes as the target fault codes, and the sample feature parameters of the positive sample and the negative sample are statistically processed respectively. Finally, the chi-square test method is used to analyze the correlation between the individual target fault codes and the sample feature parameters, to obtain the fault correlation table, so that the edge server can make the fault prediction according to the fault correlation table.

[0082] In an embodiment, determining the target fault and multiple target fault types of the target fault includes: acquiring the number of individual fault codes in the historical fault data; and filtering out the fault code with the number exceeding a set number threshold from the individual fault codes, using the fault type corresponding to the fault code as the target fault type.

[0083] In a specific implementation, for each elevator model, all the fault codes in its sample data can be processed statistically and numerically, to obtain the number of occurrences of the faults corresponding to the individual fault codes. A threshold value for the number of occurrences of the fault code can be set. The threshold value can be 50. If the number of a certain fault code is less than the threshold value, the result of the correlation analysis will be very likely to be affected by noise. This kind of fault code may not be processed, and the fault code with the number of occurrences exceeding the set number threshold can be filtered out from the individual fault codes. The fault type corresponding to the fault code is used as the target fault type.

[0084] In this embodiment, the individual fault codes in the sample data are filtered by the number of each fault code, and the fault type corresponding to the fault code with the number of occurrences exceeding the set number threshold is used as the target fault type, and the fault code with the number of occurrences less than the set number threshold is eliminated, to reduce the influence from noise on the results of the correlation analysis, and improve the accuracy of the results of the correlation analysis between the sample feature parameters and the target faults.

[0085] In an embodiment, in the above step S404, performing the correlation analysis on the sample feature parameters and the target fault codes to obtain the fault correlation table specifically includes: dividing the number of the fault codes into at least two categories; according to the number of the individual fault codes and the divided categories, performing correlation analysis on the number of the individual fault codes and the target fault codes by the chi-square test method, to obtain the correlation weight between the target fault code and the number of the individual fault codes before the occurrence of the fault type corresponding to the target fault code.

[0086] In a specific implementation, after determining the target fault type and the corresponding fault code, the number of the fault codes can be used as the sample feature parameter for the correlation analysis with the target fault. Since the chi-square test method requires that all parameters should be categorical, it is necessary to discretize the sample feature parameters firstly. Since the number of the occurrences of any fault (A, B...X, Y) within 30 minutes before the occurrence of the target fault X is mostly 0, and is seldomly not 0, the data is highly sparse. All number of the faults before the occurrence of the target faults can be divided into three categories by manual division: 0 times, 1 time, >1 time. After the category division is completed, the chi-square test method can be used to calculate the correlation between the target fault and the number of the individual fault types within 30 minutes before the target fault occurs. Since all faults are divided into three categories, the degree of freedom of the calculated results is fixed as (3-1)*(2-)=2. The results between different fault types can be compared in correlation strength through the output pvalue value.

[0087] Due to the large number of types of fault codes, and most of the fault codes have a very limited number of occurrences. For these fault codes, on one hand, it is not necessary to calculate the correlation between these fault codes and the target fault. On the other hand, the accuracy of the results of the calculated correlation is also doubtful. Therefore, in the fault part correlation analysis, only the first n (for example, the first 30) fault codes that occurs frequently can be selected for calculation, so as to improve the efficiency of the correlation analysis.

[0088] In this embodiment, the number of the fault codes is discretized into at least two categories, the correlation analysis between the number of the individual fault codes in different categories and the target fault codes is performed, to obtain the correlation weight between the number of the individual fault codes in different categories and the target fault codes, so that the edge server can obtain the fault correlation table, and then obtain the number of the individual fault codes in different categories through statistical processing of the fault data obtained in real time, so as to determine the probability of the occurrence of the target fault.

[0089] In an embodiment, in the above step S404, performing correlation analysis on the sample feature parameters and the target fault codes to obtain the fault correlation table further includes: determining multiple instantaneous indexes, performing statistical processing on the historical state data according to the preset time window length, to obtain index values of individual instantaneous indexes; discretizing the index values of the individual instantaneous indexes into at least two categories, and according to the discretized categories and the index values of the individual instantaneous indexes, performing correlation analysis on the index values of the individual instantaneous indexes and the target fault codes by the chi-square test method, to obtain the correlation weight between the target fault code and the index values of the individual instantaneous indexes before the occurrence of the fault type corresponding to the target fault code.

[0090] The instantaneous index can indicate the state of various components of the elevator within a short period of time before operation. For example, the instantaneous index can be the 40D pull-in and release time, the operating current during opening and closing the doors, and whether the elevator has previously reported other types of faults, etc.

[0091] In a specific implementation, since the purpose of performing the sample feature parameter correlation analysis is to provide a ranking based on the correlation for the state summary at the time of fault, the features of each sample can be generated with dynamically changing data. The value of the instantaneous index itself and the changing ratio in different time periods may be an important basis for determining whether the fault is true. Therefore, the instantaneous index data can be characterized by using multiple time windows. Since the purpose of the fault prediction is to predict whether the elevator will occur the predicted target fault in the next maintenance cycle, the time window (maintenance plan date - 3 days) --- > (maintenance plan + 14 days) can be

generated based on the maintenance plan of a certain elevator, and a determination of whether there is a target fault in the time window, and a determination of the fault code when there is the fault are performed.

[0092] For example, taking the instantaneous index data within 2 hours before the occurrence of the fault, and using 5 minutes as the time window length, the average value, variance, maximum and minimum values of the operating data indexes of the elevator door machine in each 5-minute window are collected. If there is no instantaneous index data in the time window, 0 is set uniformly.

[0093] Thus, the index value of the instantaneous index in each time window is obtained, and the individual instantaneous indexes in the sample features can be further discretized. Most of the instantaneous indexes are continuous values (time, current, etc.), therefore, different from the number of faults in the sample feature parameters, the instantaneous indexes are all highly sparse integer values. Therefore, the discretization method based on percentile can be used to discretize the individual instantaneous indexes. For example, if the average value of elevator door opening time from 0 to 5 minutes is intended to be discretized into 5 categories, the system will find the value of the statistical index in each row and sort it by size, and then find quantiles at 20%, 40%, 60% and 80%, divide all the "the average value of 0-5 minutes of the elevator door opening time" into 5 categories based on these 4 quantiles, and give them the corresponding index of the category to which each of them belongs. Since some instantaneous indexes may have a large number of the same value (for example, 0), in this case, there may be multiple quantiles that are actually one value. In this case, the categories with the same value can be merged automatically, that is, the number of categories obtained will be less than 5. Under the premise of ensuring that each instantaneous index is divided into at least two categories, this merging is allowed to be done automatically.

[0094] In addition, in addition to the instantaneous index having an impact on the target fault, there is also impacts of the life index data with slower data changes or the static life index data on the target fault, which are for example, the cumulative indexes of the elevator within the 24 hours before the occurrence of the target fault, such as the number of operations, the total number of 15B actions, the elevator model, and the mounting time. This type of data usually has a low direct correlation with the effectiveness of the fault when given separately. However, it may be able to provide additional supplementary information for dynamically changing data, especially in some tree algorithms, which may provide effective information gain. Therefore, the final generation form of the generated sample feature generation can be represented as shown in FIG. 6, in which the trigger type state data indicates the instantaneous index data, and the cumulative type state data may indicate the life index data.

[0095] In actual application, after statistics, it is found that some life indexes (such as the pull-in release time of 40G) have never changed, resulting in the average value, variance, etc. always being 0, which has no effect on fault filtering. In addition, there are some life indexes, which record the random state of the elevator during a certain operation, such as the floor during the elevator opening and closing the door, the number of times of reopening the light curtain, which has a low correlation with the possibility of the occurrence of the fault. Therefore, the life indexes in these two cases can be eliminated, and the finally obtained life index is used as the target life index. The average value of the target life index in each time window is calculated as the sample feature parameter.

[0096] In this embodiment, the index values of the individual instantaneous indexes are obtained through statistical processing of the historical state data according to the preset time window length. The index values of the individual instantaneous indexes are discretized into at least two categories. The correlation analysis is performed on the index values of the individual instantaneous indexes in different categories and the target fault codes, to obtain the correlation weight between the index values of the individual instantaneous indexes in different categories and the target fault codes, so that the edge server can obtain the fault correlation table, and then obtain the index values of the individual instantaneous indexes in different categories through statistical processing of the data obtained in real-time, so as to determine the probability of the occurrence of the target fault.

[0097] It should be understood that although the various steps in the flowcharts of FIGS. 3 to 5 are displayed in sequence as indicated by the arrows, these steps are not necessarily performed in sequence in the order indicated by the arrows. Unless clearly expressed herein, there is no strict restriction on the order for the performing these steps, and these steps can be performed in other orders. Moreover, at least part of the steps in FGIS. 3 to 5 may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily performed at the same time, but can be performed at different times. These sub-steps or stages are not necessarily performed sequentially, but may be performed in turn or alternately with at least a part of other steps or sub-steps or stages of other steps.

[0098] In an embodiment, a system for elevator fault prediction is provided, which includes an edge server and a central server. The edge server and the central server communicate through a network.

[0099] The central server is used to acquire historical fault data and historical state data corresponding to the historical fault data of an elevator as sample data. The historical fault data includes fault data of multiple fault types. The central server is used to input the sample data into a fault prediction model to obtain a fault correlation table of the sample feature parameters of the sample data and individual fault types. The fault correlation table records the correlation weight between individual sample feature parameters and a fault type when the fault type occurs. The central server is used to send the fault correlation table to the edge server.

[0100] The edge server is used to acquire state data and fault data of the elevator, determine current sample feature parameters according to the state data and the fault data, and determine the probability of occurrence of any fault type according to the current sample feature parameters and the fault correlation table.

[0101] In an embodiment, as shown in FIG. 7, the system further includes sub-regional nodes. The sub-regional nodes communicate with the edge server and the central server through the network, respectively. The sub-regional nodes are used to perform data statistics on the state data of the elevator acquired by multiple edge servers associated therewith, and send it to the central server.

[0102] In an embodiment, the edge server is further used to acquire the life index of the elevator operation and the index value of the life index, and determine the probability of occurrence of any fault type according to the index value of the life index, the current sample feature parameters and the fault correlation table.

[0103] In an embodiment, as shown in FIG. 8, a device for elevator fault prediction is provided, which includes a data acquisition module 802, a correlation table generation module 804, and a correlation table sending module 806.

[0104] The data acquisition module 802 is used to acquire historical fault data and historical state data corresponding to the historical fault data of the elevator as sample data. The historical fault data includes fault data of multiple fault types.

[0105] The correlation table generation module 804 is used to input the sample data into the fault prediction model to obtain the fault correlation table of the sample feature parameters of the sample data and the individual fault types. The fault correlation table records the correlation weight between the individual sample feature parameters and a fault type when the fault type occurs.

[0106] The correlation table sending module 806 is used to send the fault correlation table to the edge server. The edge server is used to acquire the state data and fault data of the elevator, determine the current sample feature parameters according to the state data and the fault data, and determine the probability of occurrence of any fault type according to the current sample feature parameters and the fault correlation table.

[0107] In an embodiment, the data acquisition module 802 is further used to determine the target fault and multiple target fault types of the target fault, and filter out the target historical fault data and the target historical state data corresponding to the target fault and the target fault type from the historical fault data and the historical state data, as the sample data.

[0108] The correlation table generation module 804 is further used to use the fault codes of the individual target fault type as the target fault codes respectively, and divide the sample data into the positive sample and the negative sample. The positive sample indicates that the fault code is the sample of the target fault code. The negative sample indicates that the fault code is not the sample of the target fault code. The correlation table generation module 804 is further used to determine multiple sample features that characterize the sample data, obtain the sample feature parameters of the individual sample features though statistical processing of the positive sample and the negative sample, and perform the correlation analysis on the sample feature parameters and the target fault codes to obtain the fault correlation table.

[0109] In an embodiment, the correlation table generation module 804 is further used to obtain the number of the individual fault codes in the historical fault data; filter out the fault code whose number of occurrences exceeds the set number threshold from the individual fault codes; and use the fault type corresponding to the fault code as the target fault type.

[0110] In an embodiment, the correlation table generation module 804 is further used to divide the number of fault codes into at least two categories; and perform the correlation analysis on the number of the individual fault codes and the target fault codes using the chi-square test method and according to the number of the individual fault codes and the divided categories, to obtain the correlation weight between the target fault code and the number of the individual fault codes before the occurrence of the fault type corresponding to the target fault code.

[0111] In an embodiment, the correlation table generation module 804 is further used to determine multiple instantaneous indexes; perform statistical processing on the historical state data according to the preset time window length to obtain the index values of the individual instantaneous indexes; discretize the index values of the individual instantaneous indexes into at least two categories; perform correlation analysis on the index values of the individual instantaneous indexes and the target fault codes using the chi-square test method and according to the discretized categories and the index values of the individual instantaneous indexes, to obtain the correlation weight between the target fault code and the index values of the individual instantaneous indexes before the occurrence of the fault type corresponding to the target fault code.

[0112] It should be noted that the device for elevator fault prediction of the present application is in one-to-one correspondence with the method for elevator fault prediction of the present application, and the technical features and beneficial effects described in the above embodiments of the elevator fault prediction method are applicable to the embodiments of the device for elevator fault prediction. The specific content can be referred to the description in the method embodiment of the present application, which will not be repeated herein, and it is hereby declared.

[0113] In addition, each module in the above device for elevator fault prediction can be implemented in whole or in part by software, hardware, and a combination thereof. The above modules may be embedded in the processor of the computer device in the form of hardware or independent of the processor of the computer device, or may be stored in the memory of the computer device in the form of software, so that the processor can call and perform the operations corresponding to the above modules.

[0114] In an embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as shown in FIG. 9. The computer device includes a processor, a memory, and a network interface connected through a system bus. The processor of the computer device is used to provide calculation and control capabilities. The memory of the computer device includes a non-transitory storage medium and an internal memory. The non-transitory storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-transitory storage medium. The database of the computer device is used to store the data generated during the elevator fault prediction process. The network interface of the computer device is used to communicate with an external terminal through a network connection. The computer program is executed by the processor to implements a method for elevator fault prediction.

[0115] Those skilled in the art can understand that the structure shown in FIG. 9 is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or less components than shown in the figure, or combining some components, or having a different arrangement of components.

[0116] In an embodiment, a computer device is further provided, which includes a memory and a processor. A computer program is stored in the memory. The processor implements the steps in the foregoing method embodiments when the computer program is executed.

[0117] In an embodiment, a computer-readable storage medium is provided, and a computer program is stored thereon. When the computer program is executed by a processor, the steps in the foregoing method embodiments are implemented.

[0118] One of ordinary skill in the art can understand that all or part of the processes in the methods in the above embodiments can be implemented by instructing relevant hardware through a computer program. The computer program can be stored in a non-transitory computer-readable storage medium. When the computer program is executed, the processes of the above method embodiments may be included. Any reference to memory, storage, database or other media used in the embodiments provided in the present application may include at least one of non-transitory and transitory memory. Non-transitory memory may include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory or the like. The transitory memory may include random access memory (RAM) or external cache memory. As an illustration and not a limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).

[0119] The technical features of the above embodiments can be combined arbitrarily. In order to simply the description, all possible combinations of the technical features in the above embodiments are not described. However, as long as there is no contradiction in the combinations of these technical features, they should be considered to be fallen into the scope defined in the present specification.

[0120] Only several implementations of the present application are illustrated in the above embodiments, and the description thereof is relatively specific and detailed, but which should not be understood as a limitation on the scope of the present application patent. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present application, several modifications and improvements can be made, which all fall within the protection scope of the present application. Therefore, the protection scope of the present application patent shall be subject to the appended claims.

Claims (10)

What is claimed is:

1. A method for elevator fault prediction, comprising,

acquiring historical fault data and historical state data corresponding to the historical fault

data of an elevator as sample data, the historical fault data comprising fault data of

multiple fault types;

inputting the sample data into a fault prediction model to obtain a fault correlation table

of sample feature parameters of the sample data and individual fault types, the fault

correlation table recording a correlation weight between individual sample feature

parameters and a fault type when the fault type occurs;

sending the fault correlation table to an edge server, wherein the edge server is configured

to obtain state data and fault data of the elevator, determine current sample feature

parameters according to the state data and the fault data, and determine a probability of

occurrence of any fault type according to the current sample feature parameters and the

fault correlation table.

2. The method according to claim 1, wherein before inputting the sample data into the

fault prediction model, the method further comprises:

determining a target fault and multiple target fault types of the target fault, and filtering

out target historical fault data and target historical state data corresponding to the target

fault and the target fault types from the historical fault data and the historical state data,

as the sample data; and

the inputting the sample data into the fault prediction model to obtain the fault correlation

table of the sample feature parameters of the sample data and the individual fault types

comprises:

using fault codes of the individual target fault types as target fault codes, and dividing the

sample data into a positive sample whose fault code is a sample of the target fault code

and a negative sample whose fault code is not a sample of the target fault code; determining multiple sample features that characterize the sample data, obtaining sample feature parameters of individual sample features through statistical processing of the positive sample and the negative sample, and performing correlation analysis on the sample feature parameters and the target fault codes to obtain the fault correlation table.

3. The method according to claim 2, wherein the determining the target fault and multiple

target fault types of the target fault comprises:

acquiring the number of individual fault codes in the historical fault data; and

filtering out a fault code with the number of occurrences exceeding a set number

threshold from the individual fault codes, using a fault type corresponding to the fault

code as the target fault type.

4. The method according to claim 2, wherein the sample feature parameters comprise

fault parameters, the fault parameters comprise the fault codes and the number of fault

codes;

the performing correlation analysis on the sample feature parameters and the target fault

codes to obtain the fault correlation table comprises:

dividing the number of the fault codes into at least two categories;

according to the number of the individual fault codes and the divided categories,

performing correlation analysis on the number of the individual fault codes and the target

fault codes by a chi-square test method, to obtain a correlation weight between the target

fault code and the number of the individual fault codes before the occurrence of a fault

type corresponding to the target fault code.

5. The method according to claim 2, wherein the sample feature parameters further

comprise state parameters, the state parameters comprise instantaneous indexes and index

values of the instantaneous indexes; the performing correlation analysis on the sample feature parameters and the target fault codes to obtain the fault correlation table further comprises: determining multiple instantaneous indexes, performing statistical processing on the historical state data according to a preset time window length, to obtain index values of individual instantaneous indexes; discretizing the index values of the individual instantaneous indexes into at least two categories, and according to the discretized categories and the index values of the individual instantaneous indexes, performing correlation analysis on the index values of the individual instantaneous indexes and the target fault codes by a chi-square test method, to obtain a correlation weight between the target fault code and the index values of the individual instantaneous indexes before the occurrence of a fault type corresponding to the target fault code.

6. A system for elevator fault prediction, comprising an edge server and a central server

that communicate through a network, wherein,

the central server is configured to acquire historical fault data and historical state data

corresponding to the historical fault data of an elevator as sample data, the historical fault

data including fault data of multiple fault types; to input the sample data into a fault

prediction model to obtain a fault correlation table of sample feature parameters of the

sample data and individual fault types, the fault correlation table recording a correlation

weight between individual sample feature parameters and a fault type when the fault type

occurs; to sending the fault correlation table to the edge server;

the edge server is configured to acquire state data and fault data of the elevator, determine

current sample feature parameters according to the state data and the fault data, and

determine a probability of occurrence of any fault type according to the current sample

feature parameters and the fault correlation table.

7. The system according to claim 6, wherein the edge server is further configured to

acquire a life index of an elevator operation and an index value of the life index, and

determine a probability of occurrence of any fault type according to the index value of

the life index, the current sample feature parameters and the fault correlation table.

8. A device for elevator fault prediction device, comprising:

a data acquisition module configured to acquire historical fault data and historical state

data corresponding to the historical fault data of an elevator as sample data, the historical

fault data comprising fault data of multiple fault types;

a correlation table generation module configured to input the sample data into a fault

prediction model to obtain a fault correlation table of sample feature parameters of the

sample data and individual fault types, the fault correlation table recording a correlation

weight between the individual sample feature parameters and a fault type when the fault

type occurs; and

a correlation table sending module configured to send the fault correlation table to an

edge server, wherein the edge server is configured to acquire state data and fault data of

the elevator, determine current sample feature parameters according to the state data and

the fault data, and determine a probability of occurrence of any fault type according to the

current sample feature parameters and the fault correlation table.

9. A computer device, comprising a memory storing a computer program and a processor,

wherein the processor implements steps of the method according to any one of claims 1

to 5 when executing the computer program.

10. A computer-readable storage medium on which a computer program is stored,

wherein when the computer program is executed by a processor, steps of the method

according to any one of claims 1 to 5 are implemented.